A moe model you say? How kawaii is it? uwu

Trinity Nano Preview: 6B parameter MoE (1B active, ~800M non-embedding), 56 layers, 128 experts with 8 active per token

Trinity Mini: 26B parameter MoE (3B active), fully post-trained reasoning model

They did pretraining on their own and are still training the large version on 2048 B300 GPUs

Looks like a less good version of qwen 30b3a which makes sense bc it is slightly smaller. If they can keep that effiency going into the large one it'll be sick.

Trinity Large [will be] a 420B parameter model with 13B active parameters. Just perfect for a large Ram pool @ q4.

> Trinity Large is currently training on 2048 B300 GPUs and will arrive in January 2026.

How long does the training take?

Excited to put this through its paces. It seems most directly comparable to GPT-OSS-20B. Comparing their numbers on the Together API: Trinity Mini is slightly less expensive ($0.045/$0.15 v $0.05/$0.20) and seems to have better latency and throughput numbers.

Moe ≠ MoE

Interesting. Always glad to see more open weight models.

I do appreciate that they openly acknowledge the areas where they followed DeepSeek's research. I wouldn't consider that a given for a US company.

Anyone tried these as a coding model yet?